"So they don't know what causes autism, but they do know what doesn't cause it."
There's nothing remarkable about that. You can test for associations between different things without having a causal factor. If vaccines caused autism it would show up in the numbers. It doesn't.